bitHound puts out features, not fires

Register or Login to like
Register or Login to like
Can government agencies be innovative?

The following is a partial transcript from a phone interview with Dan Silivestru, CEO and co-founder of—automated, open source, code quality analysis software.

Where are you from originally?

I was born in Romania, lived there for six years, don't remember any of it. Then my parents went to Israel for seven years, then moved to Montreal, Quebec, and lived there for another seven years. Then I moved to Ontario, and I'm still here.

When did you start bitHound?

We started in November 2013, and I went full-time in January 2014.

How many folks are at bitHound now?

There are nine of us. A CTO, COO, development team of four, plus staff to handle operations and HR.

What is bitHound, and what do you do?

We are centered around the concept that writing code is easy, but building resilient, remarkable software is difficult. There is much that can be told as you go along. We analyze projects from conception to today, pointing out hotspots that require attention, and suggestions on how to fix them. We track code as you move forward, so we can say if things are getting better or worse.

Something I'm proud of is a feature that showcases the dependencies that your projects have from npm and Bower, for example. It helps you understand the code that you bring into your project, and then rank it from a quality perspective. The dashboard shows you up-to-date or out-of-date status, as well as assigns you a bitHound score that is derived from Code Quality, Maintainability, and Stability. You can then pick better dependencies based on quality level. You can really dive in with bitHound.

Does bitHound support other programming languages besides JavaScript?

It is JavaScript only for now, but in the future there could be more. Rather than just the bare minimum, we think that to provide value, we've got to do a deep dive into a language. We run almost a dozen different "critics" or analyzing engines, to get "actionable insight."

(Remy: Completely understandable. Source code analysis is not what I would call a "trivial" problem...")

It is not an easy problem, it takes a lot of time and effort. We've been at it for for about a year, and it is still in closed beta.

What is it like for a bitHound user?

We strive to make the user interaction with product very simple. We think that if your software needs a manual, you're probably doing some things wrong.

The experience is simple: use OAuth for GitHub. You enable bitHound on a per-repository basis. We run our analysis, and it takes 2-20 seconds, and then we fill-in the timeline going backwards.

The idea was, on the first dashboard, you would get an "eagle-eye" view: the top five priority files, and you can expand the list further. We've had many users who are new to the concept of quality concerns such as linters, duplicate functionality, etc. So, rather than presenting an overwhelming amount of information, we present the top five most worrisome files and annotate the code with issues, so you can filter and address them. You can see on your dashboard, which dependencies are out-of-date, and we have some upcoming security analysis features in the works too.

(Remy: This sounds like it would be useful for researchers. Students in my HFOSS course at RIT have to do repository analysis as part of our "Commarch" assignment each semester.)

We have some students who use our products, and are getting introductions to professors. It seems only recently that source control is even being taught at the college level. When dealing with JavaScript, which doesn't really benefit from compiling, linters are a life-saver. The students really appreciate it. We consider these very simple things.

A big part of what we want to do behind bitHound is answer: "How can we get people to build quality code?" You have to treat your job as a craft. It is craftsmanship, and proper tooling around making software.

When did software craftsmanship become a passion for you?

It is one of those things where you get burned. You get burned in production once, twice, and then again. Then you say: "How did I get here?"

I'm self-taught from a software development process. Much of what I learned was learned "on the fly." Having gone through institutions, some left me better, some worse. The first five years of my career was focused on delivering features on time. Then I got introduced to this concept of: "If you are going to cut corners, you need to document it." When we started doing tests, though the upfront work was higher, six months later, we saw big benefits. Even as systems got more complex, you have safe-guards in place. You can go back and fix it. We were able to keep our bug-count down.

We were putting out features, rather than putting out fires.

Per feature costs, we're much lower. In the long-run, it allows your organization to move forward at a steadier, and faster, pace. Then again, at other places I've joined, they were on their second or third full rewrite. It didn't happen overnight. I didn't just wake up and say: "Test test test." It wasn't until after getting burned...

How did you get into software development?

At the University of Waterloo, honors science, then honors physics. Then I took time off to make money. While I was working at a company, I had a friend working in IT, while I was working on phones. I asked him: "IT? What do you do?" He showed me AS/400 systems and greenscreens. I asked: "How do I do that?" And the next thing I knew, I was sitting in front of the VP asking to do it.

I got a AS/400 manual, and the opportunity and big break to do that. I did that on my own time for a few weeks, and after a few months there, I said: "This is the career I want" and never looked back. I had some tremendous mentors along the way. I was there for a few years, then went into "e-business" doing consulting.

What makes a good mentor? Where and how do you find them?

The number one trait for mentors I've had, to this day, was selflessness. They were doing it for the pure joy of helping someone else develop their craft. They are not about: "I'm teaching to get something out of you for free." Obviously, they have to be knowledgeable, but you can tell more about them as a mentor by how they carry themselves.

If the first five years of my career was about: "How do I code?" After that it was about defining components that interact together. Later on, after my first consulting position, I had a new mentor, with new questions.

Dan: "We should write tests... Why?"

Mentor: "How do you know what the interface between components really looks like?"

Mentors have to be good at their craft, but you can tell a lot by the questions they ask. Listen to how they go about development and the way they ask questions.

What does your day-to-day look like?

I wrote quite a bit of code when we started, but I'm sure much of it has been rewritten. Since we announced funding in late November, it has been more about investor follow-up for me. I've matured within the code environment and in the running-a-company environment. Mostly I'm steering the ship. Day-to-day, lots of emails, some interviews like this one, working with team to set priority/strategy, and yes, I still write some code. I'm not anywhere near the critical path anymore though :P

What are your feelings on free and open source software / FOSS?

I was a co-founder of a company, tinyHippos, that we founded in 2009—which was acquired by Blackberry in 2011. One of the visions, was open sourcing the Ripple Emulator. That happened, and it was fantastic. There were only three of us that moved over, in terms of "how you run a project in FOSS."

I'm proud that they took this project, and donated it to the Apache Foundation. It sits side-by-side with PhoneGap. There was great experience in "how to foster a community" and "how do you be a BDFL?" (benevolent dictator for life). Someone who moves a project forward in a community.

We've loved FOSS throughout our career, and use it constantly at bitHound. Our analysis depends on many popular frameworks. JSHint for linting, esprima, async for callback structure, ZeroMQ for distributed parallel computing platform. You can check out our talk at JSConf last year about distributed complex computing.

Any others?

Yes! We make use of over 80 open source projects throughout our solution but a few that come to mind are d3, jquery and Polymer in production.

Where is bitHound contributing back?

Right before starting this company, Gord Tanner was the core contributor to the Apache Cordova Project; he created of the Ripple Emulator, which he donated to the Apache Foundation and is used today by Microsoft, Intel, Adobe, and over 250K developers. He unified the platform in a coherent manner, and still contributes there. For bitHound, he is a co-founder and CTO leading the technical development of services.

bitHound has simple philosophy, while we're in heavy product build, is about product, but we have come across projects in our stack that have issues. We always contribute any fixes or additions back into that project. That is our standard operating procedure. If we need to make a specific change for use that has no benefit to community at large, we don't push it, but if we fix a bug or feature, we contribute it back upstream always. That is a recommendation to any company out there. If you are going to get something for free, from someone's hard work, then if you enhance it, you should contribute it back so all can benefit. Otherwise, the community would die if everyone consumed and no one contributed.

Internally, we have components that we think will be beneficial. There will be a prominent links on our site to our GitHub.

One component is what we call "The Farm" where we have workers assigned to do work in parallel. A simple event bus really, with aggregate results coming back. We're all often dealing with the single-threaded nature of the JavaScript language, and you'll see us trying to open source that.

One thing that has happened internally with The Farm, is we've already abstracted into a separate project, to be a released. This is one of the things I've realized in my career—all of us have—just putting something on GitHub and calling it "open source" is not enough for it to take off. It must be prepared and ready for the community. That means proper README, proper docs, proper instructions for looking at project and contributing, beyond downloading and installing. Anyone can npm install but it is a different thing to make it so that contributors can understand how to augment. We will be taking our time putting our code out there, because we wanna do it right.

Any final message or parting thoughts?

Don't just write code, consider what you are doing as a craft. It takes time, and practice, and it takes time to build something that is resilient and beautiful. Open source is a great way to perfect that craft. One reason we built the dependency tool into our product was to get more people diving into code they can contribute to. Seeing other people's architectures, will expose you to better approaches.

This is sort of what spawned bitHound.

Software development is a craft, and you should be proud of it. Take the time, learn the craft, and strive to build masterpieces—the ones that gather great attention in the community. We're in this to do more than just make money, and bitHound will be free forever for open source, with no restricted features. We're huge believers in this movement. We participate, and we want to help.

This work by Remy DeCausemaker is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License. Special thanks to @itssamlowe for her contributions and edits.

At the Fedora Project Remy served as Community Action and Impact Lead, bringing more heat and light to the distro's user and contributor base.

Comments are closed.

Creative Commons LicenseThis work is licensed under a Creative Commons Attribution-Share Alike 4.0 International License.